829485 - [Camera] if the focus is drastically changed from one picture to another, you can remain on the messed up screen

Status: RESOLVED DUPLICATE of bug 830995

(Firefox OS Graveyard :: Gaia::Camera, defect)

Description

[Naoki Hirata :nhirata (please use needinfo instead of cc)]

Attachment: screenshot

## Environment :
Unagi phone, build 2013-1-10

## Repro :
1. launch camera app
2. take a picture of something far away
3. after the picture is taken move focus point to something closer
4. take a picture

## Expected :
1. preview of the second picture stays in few for a second and then goes back to camera preview

## Actual :
1. the messed up visual remains on the screen

## Note :
1. I believe there might be something odd about the autofocus timing and the capture of the picture.
2. this is permanently like this. screenshot 2 shows what happens when you tap on the screen.
3 you might have to reboot the entire phone to get out of this bad state.  if you're lucky it might clear up.  Not sure exactly in what conditions it clears.  see screenshot 3.

Comment 1

[Naoki Hirata :nhirata (please use needinfo instead of cc)]

Attachment: screenshot 2

Comment 2

[Naoki Hirata :nhirata (please use needinfo instead of cc)]

Attachment: screenshot 3

Comment 3

[Naoki Hirata :nhirata (please use needinfo instead of cc)]

Attachment: log cat

Comment 4

[Mike Habicher [:mikeh] (high bugzilla latency)]

Looking through the logcat for relevant prints:

01-11 10:55:10.692: E/QualcommCamera(659): Qint android::start_preview(camera_device*): E
01-11 10:55:10.692: V/QualcommCameraHardware(659): startPreview E
01-11 10:55:10.692: V/QualcommCameraHardware(659):  startPreview : Starting normal preview 
01-11 10:55:10.692: I/QualcommCameraHardware(659):  getBuffersAndStartPreview : E 
01-11 10:55:10.692: V/QualcommCameraHardware(659): getBuffersAndStartPreview: Calling native_window_set_buffer
  ...
01-11 10:55:10.742: D/memalloc(109): /dev/pmem_adsp: Allocated buffer base:0x43191000 size:233472 fd:105
01-11 10:55:10.742: D/memalloc(659): /dev/pmem_adsp: Mapped buffer base:0x42700000 size:233472, fd:32
01-11 10:55:10.742: V/QualcommCameraHardware(659):  Locking buffer 0 
01-11 10:55:10.742: E/QualcommCameraHardware(659):  Locked buffer 0 successfully
01-11 10:55:10.742: E/QualcommCameraHardware(659): Handle 0x425236f0, Fd passed:32, Base:0x42700000, Size 0x39000
01-11 10:55:10.742: V/QualcommCameraHardware(659): fd mmap fd 32 size 233472
01-11 10:55:10.742: E/QualcommCameraHardware(659):  Mapped Preview buffer 0
01-11 10:55:10.742: E/QualcommCameraHardware(659): Got the following from get_mem data: 0x42d46000, handle :1137019440, release : 0x4080edc1, size: 233472
01-11 10:55:10.742: E/QualcommCameraHardware(659):  getbuffersandrestartpreview deQ 32
01-11 10:55:10.742: E/QualcommCameraHardware(659): Registering buffer 0 with fd :32 with kernel
01-11 10:55:10.742: V/QualcommCameraHardware(659): use_all_chnls = 0
01-11 10:55:10.742: E/QualcommCameraHardware(659): register_buf: CbOff = 0x25800 CrOff = 0x0
01-11 10:55:10.742: V/QualcommCameraHardware(659): register_buf:  reg = 0 buffer = 0x42d46000
01-11 10:55:10.742: E/QualcommCameraHardware(659): Came back from register call to kernel
  ...
01-11 10:55:10.762: D/memalloc(109): /dev/pmem_adsp: Allocated buffer base:0x433a8000 size:233472 fd:143
01-11 10:55:10.762: D/memalloc(659): /dev/pmem_adsp: Mapped buffer base:0x430c3000 size:233472, fd:63
01-11 10:55:10.762: V/QualcommCameraHardware(659):  Locking buffer 1 
01-11 10:55:10.762: E/QualcommCameraHardware(659):  Locked buffer 1 successfully
01-11 10:55:10.762: E/QualcommCameraHardware(659): Handle 0x43cea240, Fd passed:63, Base:0x430c3000, Size 0x39000
01-11 10:55:10.762: V/QualcommCameraHardware(659): fd mmap fd 63 size 233472
01-11 10:55:10.762: E/QualcommCameraHardware(659):  Mapped Preview buffer 1
01-11 10:55:10.762: E/QualcommCameraHardware(659): Got the following from get_mem data: 0x43500000, handle :1137610816, release : 0x4080edc1, size: 233472
01-11 10:55:10.762: E/QualcommCameraHardware(659):  getbuffersandrestartpreview deQ 63
01-11 10:55:10.762: E/QualcommCameraHardware(659): Registering buffer 1 with fd :63 with kernel
01-11 10:55:10.762: V/QualcommCameraHardware(659): use_all_chnls = 0
01-11 10:55:10.762: E/QualcommCameraHardware(659): register_buf: CbOff = 0x25800 CrOff = 0x0
01-11 10:55:10.762: V/QualcommCameraHardware(659): register_buf:  reg = 0 buffer = 0x43500000
01-11 10:55:10.762: E/QualcommCameraHardware(659): Came back from register call to kernel
  ...
01-11 10:55:15.427: E/CAM_FD(659): ...config thread select timeout...
  ...
01-11 10:55:27.448: E/CAM_FD(659): ...config thread select timeout...
  ...
01-11 10:55:33.454: E/CAM_FD(659): ...config thread select timeout...

The last "config thread" messages indicate a camera process that is waiting for _something_ to happen.  In this case, it looks like the camera is waiting for the rest of the buffers to get allocated--normally there are 8 or 9 of them, but we seem to be stuck after 2.

Comment 5

[David Scravaglieri [:scravag]]

Renominate if reproducible.

Comment 6

[Naoki Hirata :nhirata (please use needinfo instead of cc)]

Found out it's easier to reproduce this issue if you take a close view of a camera and move it back from the same frame or vice versa (ie same focal point different distance)

Comment 7

[Mike Habicher [:mikeh] (high bugzilla latency)]

(In reply to Naoki Hirata :nhirata from comment #6)
>
> Found out it's easier to reproduce this issue if you take a close view of a
> camera and move it back from the same frame or vice versa (ie same focal
> point different distance)

I've tried this repeatedly on mozilla-inbound and am unable to reproduce.

Comment 8

[Naoki Hirata :nhirata (please use needinfo instead of cc)]

I can't seem to reproduce this with this morning's mozilla-inbound because I keep getting an error message after taking a picture...  I mainly revisited this because I was looking at bug 829177 with Sotaro-san.

Comment 9

[Mike Habicher [:mikeh] (high bugzilla latency)]

:nhirata, what's the error message?

Comment 10

[Sotaro Ikeda [:sotaro]]

I also see this error message in Jan/10 rom. It is Bug 828918, I think.

Comment 11

[Mike Habicher [:mikeh] (high bugzilla latency)]

nhirata, sotaro--are you able to reproduce this?  I think this problem was related to bug 830995.

Comment 12

[Sotaro Ikeda [:sotaro]]

mikeh, I cannot reproduce this anymore.

Comment 13

[Mike Habicher [:mikeh] (high bugzilla latency)]

I'm going to close this as a duplicate of bug 830995.  Feel free to reopen if it reappears.